On Fri, Apr 23, 2021 at 11:37:35AM -0400, Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > On Fri, Apr 23, 2021 at 10:08:06AM -0400, Tom Lane wrote:
> >> FWIW, I don't see any error from this, either in HEAD or the
> >> back branches.
>
> > Uh, I don't see the failure in 13 head or master, but I do see it from a
> > 13.2 tree checkout. I will try to find the post-13.2 commit cause.
>
> Oh!
>
> [ checks commit log ]
>
> I bet it's 80ca8464f.
Confirmed. The bug was added by this commit in July 2019:
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=1cff1b95a
and was included in PG 13. It was fixed by this post-PG 13.2 commit on
February 26, 2021:
https://git.postgresql.org/gitweb/?p=postgresql.git&a=commitdiff&h=80ca8464f
The commit message is very clear:
makeDependencyGraphWalker and checkWellFormedRecursionWalker
thought they could hold onto a pointer to a list's first
cons cell while the list was modified by recursive calls.
That was okay when the cons cell was actually separately
--> palloc'd ... but since commit 1cff1b95a, it's quite unsafe,
leading to core dumps or incorrect complaints of faulty
WITH nesting.
--> In the field this'd require at least a seven-deep WITH nest
to cause an issue, but enabling DEBUG_LIST_MEMORY_USAGE
allows the bug to be seen with lesser nesting depths.
Per bug #16801 from Alexander Lakhin. Back-patch to v13.
It mentions the commit it fixed, and the fact that "this'd require at
least a seven-deep WITH nest to cause an issue", which is exactly what
you saw. :-)
I am attaching this later patch in a version that will cleanly apply to
PG 13.2, in case you need it. If not, the fix will appear in 13.3 which
will be released on May 13, 2021.
--
Bruce Momjian <bruce@momjian.us> https://momjian.us
EDB https://enterprisedb.com
If only the physical world exists, free will is an illusion.
